Description du projet
Des modèles prédictifs d’apprentissage automatique pour l’évaluation des collisions et de l’aérodynamisme
Le calcul à haute performance (HPC) et l’ingénierie assistée par ordinateur (IAO) jouent un rôle décisif dans le processus de développement des véhicules. Sur l’utilisation totale du CHP dans l’automobile, environ 20 % sont captés par les simulations aérothermiques alors que jusqu’à 50 % des ressources de CHP sont exploitées aux fins des simulations de collision. Le projet UPSCALE, financé par l’UE, intègre des méthodes d’IA directement dans les logiciels et méthodes d’IAO traditionnels fondés sur la physique utilisés dans le cadre du développement des transports routiers à l’échelle mondiale. Le projet porte sur l’application des méthodes d’IA afin de réduire le temps de développement et d’accroître les performances des véhicules électrifiés, ce qui réduira les niveaux d’émission mondiaux. UPSCALE a choisi comme cas d’utilisation du projet les deux applications d’IAO les plus exigeantes en matière de HPC: la modélisation aérothermique des véhicules et la modélisation des collisions.
Objectif
UPSCALE is the first EU-project that has the specific goal to integrate artificial intelligence (AI) methods directly into traditional physics-based Computer Aided Engineering (CAE)-software and –methods. These CAE-tools are currently being used to develop road transportation not only in Europe but worldwide. The current focus of the project is to apply AI-methods to reduce the development time and increase the performance of electric vehicles (EVs) which are required by the automotive industry to reduce global emission levels. High performance computing (HPC) and CAE-software and –methods play a decisive role in vehicle development process. In order to make a significant impact on the development process, the two most HPC intensive CAE-applications have been chosen as use cases for the project: vehicle aero/thermal- and crash-modelling. When considering total automotive HPC usage, approximately 20% is used for aero/thermal simulations and up to 50% of HPC resources are utilized for crash simulations. By improving the effectiveness of these two areas, great increases in efficiency will lead to a 20% of reduction of product time to market. Other novel modelling approaches such as reduced order modelling will be coupled to the AI improved CAE-software and -methods to further reduce simulation time and ease the application of optimization tools needed to improve product quality. Through the combined effort of universities, research laboratories, European automotive OEMs, software companies and an AI-SME specialized in machine learning (ML), the UPSCALE project will provide a unique and effective environment to produce novel AI-based CAE-software solutions to improve European automotive competiveness.
Champ scientifique
Not validated
Not validated
- natural sciencescomputer and information sciencessoftware
- social sciencessocial geographytransportelectric vehicles
- engineering and technologymechanical engineeringvehicle engineeringautomotive engineering
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringcomputer hardwaresupercomputers
- natural sciencescomputer and information sciencesartificial intelligencemachine learning
Mots‑clés
Programme(s)
Régime de financement
RIA - Research and Innovation actionCoordinateur
43710 Santa Oliva
Espagne